How to Predict a Loss in a Train Using Tenser Flow

i set up four network, l s t m, using caris or tenser flow. And the my neron levels are one thousand nerons per per layer. So i have a lot of perometers actuallycrate when i run the train. And theni set my last a a layer to dance. And that's the predicted a a. Based on the ten, ten units, it will predict what the next, next future output will be.